PostHog Review

PostHog's pricing reads like a bet against its own sales team: no base fee, no seats, a free allowance on every product that resets monthly, and a billing limit you can set per product so a traffic spike cannot run up a bill you did not approve.

My view after reading the whole price list: it is the best deal in product analytics for a team under roughly 2M events a month, and a genuinely competitive one above that, as long as you watch the flag and replay meters as closely as the event meter.

This review is for developers and product engineers choosing between PostHog and the specialist tools it replaces.

I cover what the platform bundles, how the usage meters and billing limits work, what each product costs past the free allowance, a computed bill for a realistic startup, who should skip it, and the four closest alternatives priced from their own pages in October 2026.

PostHog is an open-source product platform that packages the tools a product team otherwise buys separately: product analytics, web analytics, session replay, feature flags, A/B experiments, surveys, error tracking, a data warehouse with pipelines, logs and AI observability.

It runs as a hosted service in US or EU cloud regions, and the open-source edition can be self-hosted.

The pitch is that one event stream feeds everything. A feature flag exposure becomes an experiment variant, a conversion drop in a funnel links to the session recordings that produced it, and an exception in error tracking carries the user's replay.

For a small team that means one SDK, one data model and one bill instead of a Mixpanel plus LogRocket plus LaunchDarkly plus Sentry stack.

The trade-off is depth per product. Each PostHog module competes with a specialist that does only that job; replay competes with FullStory and LogRocket, flags with LaunchDarkly, error tracking with Sentry.

PostHog wins on integration and price, and loses where a team needs the specialist's advanced features, for example LaunchDarkly's approval workflows or Sentry's performance tracing.

You install one SDK (web, mobile or server) and every product bills on its own unit: events for analytics, recordings for replay, requests for flags, responses for surveys, exceptions for error tracking, rows for the warehouse.

Each product has a free monthly allowance, and you only add a card when you want to exceed one of them. Past the allowance, volume tiers lower the unit price as usage grows, so the marginal cost of event 10 million is lower than that of event 2 million.

Two controls keep the bill predictable. A billing limit per product caps spend: when a product hits its cap, PostHog stops ingesting that product's data for the rest of the month rather than charging more. And the free plan stops at the allowance by design, so a project with no card attached can never be billed.

The paid plan, called pay-as-you-go, adds the same allowances plus six projects, seven-year data retention and email support, against one project and one-year retention on the free plan.

Platform packages sit on top of usage. Boost adds unlimited projects, white labeling, a HIPAA BAA and SSO enforcement; Scale adds SAML and priority support; Enterprise adds role-based access control, dedicated support and training.

Session recordings are kept one month on the free plan and 90 days on pay-as-you-go, extendable to a year with Boost or Scale and five years with Enterprise.

All prices below were read on PostHog's pricing page in October 2026, and competitor prices on Mixpanel pricing, Amplitude pricing, Statsig pricing and LaunchDarkly pricing the same month. PostHog prices in USD with no base fee and no seat charge.

Each product meters separately after its free allowance. Feature flag requests cost $0.0001 each from 1M to 2M, $0.000045 from 2M to 10M and $0.00001 above 50M. Survey responses are $0.10 each from 1,500 to 2,000, then $0.035 to 10,000 and $0.01 above 20,000. Error tracking exceptions are $0.00037 each from 100K to 325K, falling to $0.000115 above 10M. Data warehouse rows are $0.000015 each from 1M to 10M and $0.000001 above 1B. Group analytics and data pipelines are add-ons with their own 1M free events and per-event rates between $0.0000697 and $0.000026 depending on volume. Logs keep 14 days by default; longer retention is $0.05 per GB per month.

Cost scenario, computed from those rates: a startup at 3M events, 20K session recordings and 1.5M flag requests a month pays $50 for the second million events plus $34.30 for the third, $50 for recordings 5K to 15K plus $17.50 for the next 5K, and $50 for the half million flag requests above the allowance: about $202 a month, with no seat fees for any team size. Flags are the meter to watch: a client-side flag evaluated on every page load burns requests faster than events, and at 5M requests a month that line alone is $235. At 10M events with everything else inside the allowance, analytics costs $324.40.

Mixpanel's Growth plan prices 2M events at $185 a month and 5M at $438 on annual billing, so the same 3M-event startup lands between those figures and gets no flags or error tracking. Amplitude gives 2M events free on its Plus plan and prices above that in its calculator, so a team at 3M should check the current price on Amplitude's site. Statsig Pro is $150 a month for 5M events with unlimited flag checks and 100K replays, the strongest rival at this exact volume. LaunchDarkly's Foundation plan prices flags per service connection and per 1K client-side monthly active users (billed yearly), a different unit that favours server-side apps with few connections.

Devshot data: Toolradar tracks 83 product analytics tools as of October 2026, and 58% of them are free, freemium or open source. PostHog sits in the freemium majority, so a free allowance is table stakes in this category; what separates it is that the allowance covers six products, not one.

PostHog fits a product or engineering team that would otherwise buy three or four tools and is under a few million events a month: the whole stack is free until then, and the one-SDK, one-data-model design saves the integration work of stitching Mixpanel to LogRocket to LaunchDarkly.

It also fits a company that hates seat pricing, because 40 viewers cost nothing extra. See our feature flag platforms and error tracking tools rankings for how its modules compare one by one.

It is the wrong default for a team that needs one module at a specialist's depth: LaunchDarkly's approval workflows and release monitoring, Sentry's tracing and profiling, Amplitude's behavioral cohorts at enterprise scale.

It is also the wrong default for a company that needs SAML and SSO enforcement on day one, since those sit behind the Boost and Scale packages rather than the usage price. And a high-traffic consumer app evaluating client-side flags on every page view should model the flag meter before signing, because requests outrun events quickly.